120 Biophysical Journal Volume 66 January 1994 120-132 Influence of Cholesterol on Phospholipid Bilayers Phase Domains as Detected by Laurdan Fluorescence

Tiziana Parasassi,* Massimo Di Stefano,* Marianna Loiero,* Giampietro Ravagnan,* and Enrico Grattont *Istituto di Medicina Sperimentale, Consiglio Nazionale Ricerche, 00137 Rome, Italy, and tLaboratory for Fluorescence Dynamics, University of Illinois at Urbana-Champaign, Urbana, Illinois, USA

ABSTRACT Coexisting gel and liquid-crystalline phospholipid phase domains can be observed in synthetic phospholipid vesicles during the transition from one phase to the other and, in vesicles of mixed phospholipids, at intermediate temperatures between the transitions of the different phospholipids. The presence of cholesterol perturbs the dynamic properties of both phases to such an extent as to prevent the detection of coexisting phases. 6-Lauroyl-2-dimethylaminonaphthalene (Laurdan) fluorescence offers the unique advantage of well resolvable spectral parameters in the two phospholipid phases that can be used for the detection and quantitation of coexisting gel and liquid-crystalline domains. From Laurdan fluorescence excitation and emission spectra, the generalized polarization spectra and values were calculated. By the generalized polarization phos- pholipid phase domain coexistence can be detected, and each phase can be quantitated. In the same phospholipid vesicles where without cholesterol domain coexistence can be detected, above 15 mol % and, remarkably, at physiological cholesterol concentrations, .30 mol %, no separate Laurdan fluorescence signals characteristic of distinct domains can be observed. Consequences of our results on the possible size and dynamics of phospholipid phase domains and their biological relevance are discussed.

INTRODUCTION Together with phospholipids, cholesterol is ubiquitous in nance (ESR) (Subczynski et al., 1990; Sankaram and animal cell membranes and is one of the major modifiers of Thompson, 1990b), calorimetry (Vist and Davis, 1990; the phospholipid membrane structure and dynamics. In most Tampe et al., 1991; Keough et al., 1989), dilatometry biological membranes cholesterol concentration is relatively (Melchior et al., 1980), small angle neutron scattering high, generally .30 mol %. Phospholipids in biological (Mortensen et al., 1988), and fluorescence (van Langen et al., membranes have a complex composition, including differ- 1989; Schroeder et al., 1991; Ben-Yasar and Barenholz, ences in length and in unsaturation of their acyl residues and 1989; Nemecz and Shroeder, 1988). Abrupt variations of in the type of polar heads. These chemical differences cause several structural and dynamical parameters have been found different dynamic properties and, in synthetic phospholipid at critical cholesterol concentrations. These variations have membranes, they were shown to favor the appearance of been interpreted as arising from specific complexes between segregated coexisting domains having different phase states lipids and cholesterol in the molar ratios 3:1, 2:1, and 1:1. and dynamical properties, close to the pure gel and liquid- Ipsen et al. (1987) proposed an alternative and more satis- crystalline phases (Shimshick and McConnel, 1973; factory interpretation. Their model does not require specific Parasassi et al., 1984; Parasassi et al., 1993). The gel phase complexes between phospholipids and cholesterol. The im- domain is composed of the with higher tran- phospholipid portant ingredient of this model is the different interaction sition temperature (Tm) in the presence of the phospholipid free energy between all possible pair combinations of chain- with lower Tm at concentration s30 mol %. The liquid- crystalline phase domain has a complementary composition. ordered, chain-disordered, and cholesterol. Phase diagrams Thus, the properties of the two phase domains are close to of cholesterol in dimyristoylphosphatidylcholine (DMPC) the pure phases, modified by the presence of the other phos- (Tampe et al., 1991), in deuterated dipalmitoylphosphatidyl- pholipid, but still resolvable (Parasassi et al., 1993). Impor- choline (DPPC) vesicles (Vist and Davis, 1990), and partial tant questions remain about the influence of cholesterol on phase diagrams in DPPC and sphingomyelin (Sankaram and the dynamic properties and on the coexistence of phospho- Thompson, 1990b) have been constructed, based on NMR, lipid phases. Cholesterol effect on phospholipid bilayers has ESR, differential scanning calorimetry, fluorescence, and been studied by a variety of spectroscopic techniques, in- theoretical studies as shown by Ipsen et al. (1987). As a first cluding nuclear magnetic resonance (NMR) (Vist and Davis, approximation, at concentrations above 6 mol %, cholesterol 1990; Sankaram and Thompson, 1990a), electron spin reso- has a disordering effect on the gel phase and an ordering effect on the liquid-crystalline phase (Vist and Davis, 1990). Specifically, in the gel phase, lateral diffusion and axial ro- tation Received for publication 9 August 1993 and in final form 29 September are increased by cholesterol, while in the liquid- 1993. crystalline phase axial rotational motion is decreased Address reprint requests to Dr. Tiziana Parasassi, Istituto di Medicina Speri- (Sankaram and Thompson, 1990a, Rubenstein et al., 1979). mentale, Consiglio Nazionale Ricerche, Viale Marx 15, 00137 Rome, Italy. Above a cholesterol concentration of 30 mol %, in the liquid- X 1994 by the Biophysical Society crystalline phase, lateral diffusion is decreased (Ipsen et al., 0006-3495/94/01/120/13 $2.00 1987; Mouritsen, 1991). From the phase diagrams, apart Parasassi et al. Cholesterol Suppresses Phospholipid Domains 121 from the pure gel and liquid-crystalline phases observed in coexistence of different dynamical properties is expected. the absence of cholesterol, four main regions have been clas- However, the difficulty of applying fluorescence spectros- sified by Vist and Davis (1990): (a) at low temperatures and copy to this problem resides in finding a fluorescent probe above 6 mol % cholesterol, the coexistence of gel and liquid- that displays characteristic distinct properties in the two ordered phases can be observed; (b) at high temperatures and phases. In the present study we used the spectral sensitivity at low cholesterol concentrations, the liquid-disordered of the fluorescent probe 2-dimethylamino-6-lauroylnaph- phase is observed; (c) at high temperatures and at high cho- thalene (Laurdan) to investigate the influence of cholesterol lesterol concentrations, the liquid-ordered phase is observed; on the structure and dynamics of the two phases of phos- (d) at high temperatures and intermediate cholesterol con- pholipid bilayers, with the specific purpose of further ex- centrations, the coexistence of liquid-ordered and liquid- ploring the rapid mixing hypothesis. disordered phases occurs. Following the above classification, the coexistence of gel and liquid-crystalline phases can only be observed at cholesterol concentrations below 6 mol % in LAURDAN FLUORESCENCE PROPERTIES a narrow temperature interval, about 1°C wide (Vist and In this section we review the fluorescence properties of Davis, 1990; Tampe et al., 1991) corresponding to the tran- Laurdan that are relevant to the present study of domain sition temperature of the phospholipid. coexistence and dynamics. We have shown that the fluores- The major feature derived from the above results is that cence properties of Laurdan are extremely sensitive to the because of the combined effects on the lateral diffusion and polarity of the environment of its fluorescent moiety on the axial rotational motion of phospholipids, and because (Parasassi et al., 1991). Laurdan emission spectrum is bluer of the appearance of the new liquid-disordered and liquid- in apolar solvents. In addition to the relatively small emission ordered phases, the transition of phospholipids from the gel shift due to changes in polarity, in phospholipid vesicles to the liquid-crystalline phase progressively disappears by there is a 50-nm emission red spectral shift of Laurdan fluo- increasing cholesterol concentration. As a result, no calori- rescence when passing from the gel to the liquid-crystalline metric transition can be detected in phospholipids at cho- phase (Parasassi et al., 1990). This large spectral shift is due lesterol concentration >17 mol % (Vist and Davis, 1990; to dipolar relaxation processes that occur in the liquid- Keough et al, 1989). At cholesterol concentrations above 6 crystalline phase. Since the amount of spectral shift depends mol %, the phospholipid pretransition is suppressed (Vist and on the rate of dipolar relaxation, a small modification of the Davis, 1990). dynamical properties of the membrane can cause very large The regions in which two phases coexist have been care- effects on Laurdan emission properties. Also, Laurdan ex- fully analyzed by a number of techniques. The basic idea is citation spectrum is different in the two phospholipid phases that in the region of phase coexistence, two separate signals (Parasassi et al., 1990). The excitation maximum in gel phase indicating two sets of different spectroscopic properties phospholipids is at about 390 nm, while in the liquid- should be detected. In their NMR study, Vist and Davis crystalline phase the excitation maximum is at about 360 nm. (1990) proved that it is possible to resolve the NMR spectra Because of this difference in excitation, a partial photos- of DPPC at various cholesterol concentrations using the lin- election of Laurdan molecules surrounded by the gel phase ear superposition of two components, one characteristic of with respect to those in the liquid-crystalline environment the gel phase with low cholesterol concentration and one is possible. For analogy with the concept of photoselection characteristic of the liquid-crystalline phase formed at high that can be performed using polarized excitation, spec- cholesterol concentration. However, this decomposition can troscopic properties of Laurdan have been described be performed only at temperatures below the major phase (Parasassi et al., 1990 and 1991) by the generalized polar- transition of DPPC but not at high temperatures. The reason ization (GP) defined as for this failure is attributed to a rapid mixing (in the NMR characteristic time scale) of the signals from the two phases GP = (Ig - I)/(Ig + I) (1) at high temperatures. A similar conclusion was reached by where Ig and I, are the fluorescence intensities at the maxi- Sankaram and Thompson (1990b) using ESR in various mum emission in phospholipids in the gel and in the liquid- phospholipids and sphingomyelin. Also in this case there is crystalline phase, respectively (excitation GP). In addition, not direct observation of two separate signals characteristic the GP value can be obtained using the fluorescence inten- of two distinct phases. An average signal is instead observed. sities at the maximum excitation in phospholipids in the gel The failure to observe two separate spectroscopic properties and in the liquid-crystalline phase (emission GP). was again attributed to rapid mixing-rapid as compared to A further peculiar feature of Laurdan spectroscopy con- the ESR characteristic time scale. This characteristic rate is cerns the red band of the excitation spectrum, centered at on the order of 5 x 107 s-1. about 390 nm. This second excitation band is absent in apolar In principle, fluorescence spectroscopy can explore a solvents, appears in polar solvent capable of hydrogen bond- much faster rate, in the range of 108-109 s-1. Therefore, fluo- ing, and is quite intense in phospholipid vesicles (Parasassi rescence spectroscopy offers the opportunity ofverifying the et al., 1991). In phospholipids, the fluorescent moiety of rapid mixing hypothesis as the reason for the failure of the Laurdan is located at the level of the glycerol backbone, observation of distinct spectroscopic properties where the where a few partially immobilized water molecules may be 122 Biophysical Journal Volume 66 January 1994 present. Thus, the red excitation band seems to originate Fluorescence measurements from the stabilization of Laurdan ground state by a local Laurdan excitation and emission spectra in the lipid vesicles were obtained polar environment. Moreover, this band is populated by using a GREG 200 fluorometer, equipped with photon-counting electronics Laurdan molecules with a favorable orientation of sur- (PX01, ISS Inc., Champaign, IL) and the accompanying software. The fluor- rounding dipoles, already relaxed, but is associated to gel ometer uses a xenon arc lamp as the light source. Monochromator bandpass phase, unrelaxed phospholipids. By selectively exciting was 4 nm. The spectra were only corrected for lamp intensity variations. For the DLPC-DPPC samples, the corresponding blank spectra were acquired this red band, the resulting shape of Laurdan emission de- and subtracted from the fluorescence spectra. The emission GP spectra were pends on the balance between the amount of relaxation, constructed by calculating the GP value for each emission wavelength as i.e., the phospholipid phase state, and the local polarity of follows: the Laurdan environment. GP = (I410 -I340)/(I410 + I340) (2) The GP value offers several advantages for the measure- ment and analysis of Laurdan fluorescence data. In phos- where '410 and I340 are the intensities at each emission wavelength, from 425 nm to 550 nm, obtained using fixed excitation wavelength of 410 nm and pholipids of different polar heads and of various acyl resi- 340 nm, respectively. The excitation GP spectra were constructed in a simi- dues, and at pH values varying from 4 to 10, the GP has a lar way from the excitation spectra using: unique characteristic value for the gel and for the liquid- crystalline phase (Parasassi et al., 1991). The excitation GP GP = (I440 -I490)1(I440 + I490) (3) value in the gel phase is high, of about 0.6, while in the where I440 and I490 are the intensities at each excitation wavelength, from liquid-crystalline phase is low and negative, of about -0.3. 320 nm to 420 nm, obtained using fixed emission wavelength of 440 nm and 490 nm, respectively. For the calculation of GP, fixed excitation and Moreover, the property of additivity of the GP can be used emission wavelengths were chosen as previously discussed (Parasassi et al., to quantitate coexisting phases (Parasassi et al., 1991). 1990). In particular, in the case of the calculation of emission GP, excitation Laurdan GP value can be obtained both from emission and wavelengths of 340 nm and 410 nm do not correspond to the maximum from excitation spectra, providing independent information excitation of the liquid-crystalline and of the gel phase, respectively. They and on the of the were chosen as convenient extremes of the excitation spectrum to improve on the dipolar relaxation process polarity the photoselection of the two Laurdan populations. Temperature was con- environment. The particular wavelength behavior of the GP trolled to ± 0.1°C by a water circulating bath. value gives information on the coexistence of different Oxygen-quenching experiments were performed using a photon- phases and on their interconversion (Parasassi et al., 1993). counting fluorometer, model GREG PC (ISS Inc.) equipped with a xenon When Laurdan is used in cell membranes, where choles- arc lamp and an oxygen pressure cell described in Lakowicz and Weber (1973) holding a 2 X 2 cm cuvette and thermostated at 20°C by a circulating terol and a variety of lipids are present, a blue shift of its water bath. During measurements, samples were continuously stirred. The emission and excitation spectra is observed (Parasassi measurements in the absence of oxygen were performed in N2 atmosphere. et al., 1992). A similar blue shift was observed in DMPC After the application of each value of oxygen pressure, samples were al- vesicles after the addition of 30 mol % cholesterol. In cell lowed to equilibrate for 45 min in the dark with continuous stirring. membranes Laurdan GP value is relatively high, and in The three-dimensional plots reported in Figs. 9-11 were created using the software Axum from TriMetrix Inc. The underlying grid was obtained DMPC vesicles the GP value is higher compared with the from the same data of the plots, using the three-dimensional spline surface value obtained in the absence of cholesterol at all tempera- option of the Axum program. tures (Parasassi et al., 1992). With the purpose of 1) verifying the existence and kinetics RESULTS of the mixing between separate domains in phospholipid vesicles and 2) investigating the origin of the blue spectral We have performed four different kinds of experiments on shift and of the relatively high GP value observed in cell phospholipid vesicles labeled with Laurdan with the aim of membranes, we explored the influence of cholesterol on studying the coexistence of domains in the presence of cho- Laurdan spectroscopic properties. The study was performed lesterol: 1) effect of different cholesterol concentration in in various phospholipid vesicles with cholesterol concentra- phospholipid vesicles on Laurdan excitation and emission tions up to 60 mol %. spectra to identify the effect of polarity changes; 2) tem- perature dependence of Laurdan GP value in phospholipid vesicles and in the presence of different cholesterol concen- MATERIALS AND METHODS trations to assess the effect of cholesterol on Laurdan dipolar relaxation; 3) wavelength dependence of Laurdan excitation Laurdan-labeled liposomes and emission GP spectra in phospholipid vesicles with vari- ous cholesterol concentrations to detect the coexistence of Multilamellar phospholipid vesicles were prepared using conventional methods. The appropriate amounts of phospholipids in chloroform solution phase domains; 4) a surface for each of the phospholipids (dilauroyl-, dimyristoyl-, and dipalmitoyl-phosphatidylcholine; DLPC, used, built with the purpose of separating the influence of DMPC and DPPC, respectively) (Avanti Polar Lipids, Inc., Alabaster, AL), cholesterol on Laurdan dipolar relaxation from the polarity cholesterol (Sigma Chemical Co., St. Louis, MO), and Laurdan (Molecular changes. Probes Inc., Eugene, OR) were mixed. The solvent was evaporated by ni- trogen flow. The dried samples were resuspended in phosphate-buffered saline solution (ICN Biomedicals, Irvine, CA), heated to 700C, and vor- Laurdan excitation and emission spectra texed. All samples were prepared in red light and used immediately after preparation. The final lipids and probe concentrations were 0.3 mM and 0.3 Laurdan excitation and emission spectra were measured in AM, respectively. Chloroform was spectroscopic grade. DLPC, DMPC, and in an equimolar DLPC-DPPC mixture, Parasassi et al. Cholesterol Suppresses Phospholipid Domains 123 from 1°C to 65°C, at cholesterol concentrations from 0 to the gel phase, so that at higher cholesterol concentrations the 60 mol %. As representative examples, Laurdan excitation phase transition is no longer detectable. At low temperatures, and emission spectra obtained in multilamellar vesicles corresponding to the DMPC gel phase, the addition of cho- composed of DMPC or of an equimolar mixture of DLPC lesterol causes an increase of the excitation GP value and a and DPPC with no cholesterol and with 30 mol % choles- small decrease of the emission GP value. terol at 1°C are reported in Fig. 1, A and C, respectively. Excitation and emission GP values versus temperature Both excitation and emission spectra are blue shifted in the were calculated from Laurdan spectra obtained in samples presence of cholesterol. At 65°C the emission blue shift in composed of an equimolar mixture of DLPC and DPPC with the presence of cholesterol is quite relevant (Fig. 1, B and increasing cholesterol concentrations, from 5 to 60 mol %, D), while the excitation spectrum shows a red shift accom- and some representative curves are reported in Fig. 2, C and panied by the increase in intensity of the red band of exci- D. For these samples, blank spectra were acquired and sub- tation at about 390 nm. tracted from the excitation and emission Laurdan spectra. The effect of blank subtraction is relatively small (GP value Temperature dependence of Laurdan GP difference of about 3%) and the major effect of the blank subtraction was observed at the edges of both excitation and The excitation GP values obtained in DMPC vesicles with emission GP spectra. For the construction of Fig. 2, C and cholesterol concentrations from 5 to 30 mol % using an ex- D, an excitation wavelength of 370 nm rather than 340 nm citation wavelength of 340 nm and emission wavelengths of and an emission wavelength of 490 nm rather than 440 nm 440 and 490 nm versus temperature are reported in Fig. 2A. were chosen as the wavelengths less affected by the blank In Fig. 2 B the emission GP values obtained in the same subtraction. Similar to the results obtained using DMPC samples using an emission wavelength of 440 nm and ex- vesicles, by adding cholesterol to this phospholipid mixture, citation wavelengths of 410 and 340 nm are reported versus both excitation and emission GP values increase. At tem- temperature. The temperature interval of the phase transition peratures above 25°C the increase of the GP value is par- is progressively broader and shifted at higher temperature by ticularly relevant, and the effect on the GP variation with increasing the cholesterol concentration in DMPC vesicles as temperature is smoothed. Similar results (data not shown) detected by both the excitation and emission GP values. In were also obtained for DLPC, where the GP follows the high the temperature range of DMPC liquid-crystalline phase, the temperature behavior observed with DMPC. increase of cholesterol concentration produces an increase of In Fig. 3 the difference between the excitation (Fig. 3, A both excitation and emission GP values to the GP values of and C) and the emission (Fig. 3, B and D) GP values obtained

Excitation Emission Excitation Emission C 1.000

0.750

0.500 a) N N ._ 0.250 \\ ~~~~~co E E C- o 0.000 L 0.000 l 320 378 435 493 550 c 1- 320 378 435 493 550 D 1.000 1.000 / / 650C a) BD 65 - 0.750 C 0.750

0.500 0.500

0.250 0.250

0.000 0.000 320 378 435 493 550 Wavelength (nm) Wav elength (nm)

FIGURE 1 Laurdan normalized excitation and emission spectra obtained in multilamellar vesicles composed of DMPC (A and B) and of an equimolar DLPC-DPPC mixture (C and D) with no cholesterol ( ) and with 30 mol % cholesterol (- -) at 1°C (A and C) and at 65°C (B and D). 124 Biophysical Journal Volume 66 January 1994

0.4 B

0.1 0L 0L CD O E ai) -0.2 030

5 10 20 30 40 50 60 7 D Temperature ( C) Temperature (C)

a- C!,a. E

20 30 40 50 20 30 40 50 TemDerature ( C) TemDerature (C)

FIGURE 2 Laurdan excitation (A and C) and emission (B and D) GP values versus temperature obtained in multilamellar vesicles at various cholesterol concentrations. The percent values represent mol % of cholesterol in phospholipids. The vesicle composition was DMPC (A and B) and equimolar DLPC- DPPC (C and D). at various cholesterol concentrations and without cholesterol trations and at different temperatures are reported in Fig. 4. is reported for DMPC vesicles (Fig. 3, A and B) and for Both excitation and emission GP at 1°C do not show a rel- vesicles composed of equimolar DLPC-DPPC (Fig. 3, C and evant wavelength dependence, and this behavior is not af- D). Fig. 3 shows that (a) major variations of the emission GP fected by cholesterol addition. This wavelength indepen- value occur at a temperature a few degrees lower with respect dence of GP value is characteristic ofthe gel phase (Parasassi to variations of the excitation GP value; (b) the increase of et al., 1993). At 65°C, excitation GP spectra show decreasing both excitation and emission GP values with the increase of values as the excitation wavelength increases up to 10 mol cholesterol concentration is particularly relevant in the % cholesterol (Fig. 4 A), while emission GP spectra show liquid-crystalline phase; (c) when cholesterol is added, the increasing values as the emission wavelength increases (Fig. excitation GP values generally show larger variations than 4 B). At high cholesterol concentration, the slope due to the the emission GP values, i.e., the variation of the emission wavelength dependence is decreased. This behavior of the spectra are larger than those of the excitation spectra; (d) for GP versus wavelength observed at 65°C is characteristic of the DMPC sample, the variation of emission GP at low tem- phospholipids in the liquid-crystalline phase (Parasassi et al., perature is negative; and (e) a partial saturation effect can be 1993). The temperature at which the behavior of GP versus observed at high cholesterol concentrations. In DMPC wavelength changes from that characteristic of the gel to that vesicles, small variations of emission and excitation GP val- characteristic of the liquid-crystalline phase depends on cho- ues are observed by increasing cholesterol concentration lesterol concentration and increases with the increase of cho- above 20 mol % (Fig. 3, A and B). In the equimolar DLPC- lesterol concentration, in agreement with the shift of the tran- DPPC sample, the excitation GP value shows larger varia- sition temperature reported in Figs. 2 and 3. tions for cholesterol addition up to 30 mol % and the emission Similarly, in vesicles composed of an equimolar mixture GP value for cholesterol addition up to 35 mol % (Fig. 3, C of DLPC and DPPC, at low and at high temperatures, the and D). A similar behavior was observed for DLPC vesicles wavelength dependence of excitation and emission GP is not at different cholesterol concentrations (data not shown). affected by cholesterol addition up to 30 mol % (Fig. 5). At low temperature, the effect of cholesterol only consists of a Wavelength dependence of Laurdan GP spectra slight increase of the GP value, without affecting the wave- length independence (Fig. 5). At high temperature, the ad- Some representative Laurdan excitation and emission GP dition of cholesterol causes a relevant increase of both ab- spectra in DMPC vesicles with various cholesterol concen- solute excitation and emission GP values (Fig. 5, A and B, Parasassi et al. Cholesterol Suppresses Phospholipid Domains 125

a- 0. x a- U) E c CD a) a)c C 0 U) Ca) c]a) a)

0 0

Temperature (C) Temperature (C)

a- a. 0 co x U1) E C U) C U1)C.) .) C C a) a) U) 0

30 40 50 Temperature (C) Temperature (C)

FIGURE 3 Difference between Laurdan excitation (A and C) and emission (B and D) GP values obtained in phospholipid multilamellar vesicles with various cholesterol concentrations and without cholesterol as a function of temperature. The percent values represent mol % of cholesterol in phospholipids. The vesicle composition was DMPC (A and B) and eqimolar DLPC-DPPC (C and D). respectively, and Figs. 2 and 3), with little change of the lesterol, the behavior of GP values changes from the absence wavelength dependence. Above 30 mol % cholesterol, very of an appreciable wavelength dependence observed at low little wavelength dependence ofboth emission and excitation temperatures, to the wavelength dependence typical of the GP values can be observed also at high temperatures. At liquid-crystalline phase. This last behavior is observed re- intermediate temperatures, particularly between 10 and gardless of the absolute GP values. 30°C, the major variation caused by cholesterol concerns the In DLPC vesicles, the wavelength dependence of the GP change of the typical wavelength dependence of GP spectra, value typical of the liquid-crystalline phase is not modified which is observed when coexisting domains of gel and by increasing cholesterol concentration (data not shown). liquid-crystalline phases are present. In Fig. 6 the GP spectra The absolute excitation and emission GP values increase obtained in the equimolar DLPC-DPPC mixture at 25°C are with the addition of cholesterol, but, regardless to cholesterol shown. In the absence of cholesterol, the equimolar mixture concentration, the excitation GP spectra show decreasing of DLPC and DPPC at temperatures from about 10°C to values with the increase of excitation wavelength, and the about 30°C gives increasing values of the excitation GP spec- emission GP spectra show increasing values with the in- trum and decreasing values ofthe emission GP spectrum with crease of emission wavelength. This behavior is similar to the increase of excitation and emission wavelength, respec- that reported in Fig. 4, A and B, for DMPC at high tem- tively. Of note is the fact that this wavelength dependence is perature. Similarly to the DLPC-DPPC mixture, above 30 opposite to that observed in the pure liquid-crystalline phase. mol % cholesterol very little variation of GP values as a Indeed, without cholesterol, the type of wavelength depen- function of wavelength was observed (data not shown). dence of emission and excitation GP values can be used to To verify the possibility that Laurdan preferentially as- ascertain the coexistence of phospholipid phase domains sociates with cholesterol or that the probe will partition in (Parasassi et al., 1993). This behavior can still be observed possible domains with high cholesterol concentration, the in the phospholipid mixture after cholesterol addition up to determination of the partition coefficient of the probe be- 10 mol % for the excitation GP spectrum (Fig. 6) and up to tween phospholipids and cholesterol must be obtained. The 15 mol % for the emission GP spectrum (data not shown). equipartition of Laurdan between the two phospholipid At these cholesterol concentrations, the temperature range at phases has been already demonstrated (Parasassi et al., which the wavelength dependence typical of coexisting do- 1993). Excitation GP values obtained using excitation wave- mains can be observed is restricted. Above 15 mol % cho- lengths of 340 or 410 nm and emission wavelengths of 440 126 Biophysical Journal Volume 66 January 1994 A A 0.700 10%, 20%, 30% 0.700 10-60% _ _ _---i-_ _ 100c 0% 0.375 0.400 30% 60% GP 650C GP 0.100 20% 0.050 ______30% ______.- 20%/O - *------10% 650 C %1Q°/O -0.200 -0.275 0% - -l 0%

-0.500 -0.600 - - 320 345 370 395 420 320 340 360 380 400 B 0.350 2% 1-°C Z.- ~~~~~~~~~~~~~~:- B _~ _ - ______r_.~- 0.300

0.138 1 -15% --,- -0 :< _1-- =g GP 0.050 -0.075

30% _ GP 650C 60% -0.200 650C -0.287 - 25%-/o- 0%-30% -0.450 -0.500 00/I 425 446 468 489 510 Wavelength (nm) -0.700 L I I 42'8 459 489 520 550 FIGURE 4 Laurdan excitation (A) and emission (B) GP spectra obtained in DMPC vesicles at various cholesterol concentrations, at 1°C and at 65°C. Wav elength (nm) The percent values represent mol % of cholesterol in phospholipids. FIGURE 5 Laurdan excitation (A) and emission (B) GP spectra obtained in the equimolar DLPC-DPPC mixture at various cholesterol concentra- tions, at 1°C and at 65°C. The percent values represent mol % of cholesterol and 490 nm in DLPC vesicles at 20°C with cholesterol con- in phospholipids. centrations varying from 1 to 55 mol % are reported in Fig. 7. In Fig. 7 A, the straight line segments from 0 to 30 mol obtained using phospholipids in a single phase. In the % cholesterol, obtained by joining the GP values at 0 and 30 equimolar DLPC-DPPC mixture, at temperature where the mol % cholesterol, represent the GP values expected from a coexistence of separate domains of different phases occurs, mixture of DLPC and cholesterol in which Laurdan shows the Stern-Volmer plot was biphasic. By adding 10 mol % ideal homogeneous partitioning. The small deviation of the cholesterol to vesicles in the gel phase, a biphasic Stern- experimental GP values from the straight line is to be at- Volmer plot was also obtained (Parasassi and Gratton, tributed to dipolar relaxation and to interconversion between 1992). Here we performed oxygen-quenching experiments domains, as previously reported (Parasassi et al., 1993). A of Laurdan fluorescence in DLPC vesicles at 20°C (Fig. saturation can be observed in the excitation GP above 30 mol 8). In this pure liquid-crystalline phase the Stern-Volmer % cholesterol (Fig. 7A) and in the emission GP above 40 mol plot is linear, as already reported (Parasassi and Gratton, % cholesterol (Fig. 7B). From preliminary experiments (data 1992). By adding cholesterol up to 55 mol %, no deviation not shown) performed using a dynamic fluorescence tech- from linearity was observed (Fig. 8), although by increas- nique for the measurement of time-resolved emission spectra ing cholesterol concentration the quenching constants are of Laurdan in DLPC at various cholesterol concentrations, quite different, by more than a factor of 2 at 55 mol % we observed a modification of Laurdan relaxation kinetics cholesterol. due to the presence of cholesterol. Nevertheless, no specific relaxation kinetics due to the presence of cholesterol and Influence of cholesterol on bilayer polarity and indicative of a preferential partitioning of Laurdan in sepa- on Laurdan dipolar relaxation rated cholesterol domains was observed. From previous oxygen-quenching experiments of Laurdan fluorescence Using emission and excitation GP values and the tempera- (Parasassi and Gratton, 1992), a linear Stern-Volmer plot was ture of their measurement obtained with each phospholipid Parasassi et al. Cholesterol Suppresses Phospholipid Domains 127

Excitation Emission 0.700 30%

0.462 10% GP 0.225 0 30% -0.013 _ 10 20%f -0.250L 000 320 378 435 493 550 Wavelength (nm) 400 Oxvqen (psi) FIGURE 6 Laurdan excitation and emission GP spectra obtained in the equimolar DLPC-DPPC mixture at various cholesterol concentrations at FIGURE 8 Stern-Volmer plot obtained by oxygen-quenching experi- 25°C. The percent values represent mol % of cholesterol in phospholipids. ments of Laurdan fluorescence in vesicles composed of DLPC without cho- lesterol and with various cholesterol concentrations at 20°C. Percent values represent mol % of cholesterol in DLPC. sample with different cholesterol concentrations, a three- dimensional surface was constructed. In Fig. 9 the surface obtained with DMPC and cholesterol concentrations of 0, the surface obtained with DLPC and cholesterol concentra- 3, 5, 10, 15, 20, 25, and 30 mol % are shown. In Fig. 10A, tions of 0, 3, 5, 10, 15, 20, 25, 30, 35, 40, 45, and 50 mol the surface obtained with the equimolar DLPC-DPPC mix- % is shown. On each surface, the individual three- ture and cholesterol concentrations of 0, 3, 5, 10, 15, 20, dimensional curves of samples at each cholesterol concen- 25, 30, 35, 40, 45, and 50 mol % is shown. In Fig. 11 A, tration are superimposed. To better display the area of the high cholesterol concentration, in Fig. 10 B the surface ob- tained with the equimolar DLPC-DPPC mixture in the presence of 25, 30, 35, 40, 45, 50, 55, and 60 mol % cho- lesterol is reported, and in Fig. 11 B the surface obtained using DLPC and cholesterol concentrations of 25, 30, 35, 40, 45, 50, 55, and 60 mol % is reported. C- (2,

mol% cholesterol in DLPC

., %.fv7 B ._-- -0.1

-0.2 EL -0.3- em=4500

-0.4' / m=450 FIGURE 9 Three-dimensional surface obtained from Laurdan excitation -0.54i and emission GP values and from the temperature of their measurement in 0 5 10 15 20 25 30 35 40 45 50 55 DMPC vesicles with cholesterol concentrations of 0, 3, 5, 10, 15, 20, 25, mol% cholesterol in DLPC and 30 mol %. The excitation GP values were measured using excitation at 370 nm and the emission GP values using emission at 490 nm. The surface FIGURE 7 Laurdan excitation (A) and emission (B) GP values in DLPC was generated using all the data by the three-dimensional spline option of vesicles with increasing cholesterol concentrations at 20°C. GP values were the Axum software, as reported in Materials and Methods. The individual obtained using excitation wavelengths of 340 and 410 nm (A) and emission three-dimensional curves of samples at each cholesterol concentration are wavelengths of 450 and 500 nm (B). superimposed. 128 Biophysical Journal Volume 66 January 1994

In general, by adding cholesterol to phospholipids, DISCUSSION Laurdan excitation GP value varies by a different amount To understand at a molecular level the effect of cholesterol with respect to the variation of emission GP value; i.e., for on phospholipid bilayers, we discuss first how the properties each phospholipid, the three-dimensional plots obtained at of Laurdan are influenced by the nature and dynamics of its various cholesterol concentrations do not overlap. local surroundings. Laurdan spectral properties are depend- In these surfaces different areas can be characterized. High ent on at least two major factors: 1) the polarity of the en- values of both excitation and emission GP at low tempera- vironment; and 2) the rate of relaxation of molecules or mo- tures define an area of gel-like phase where little relaxation lecular residues that can reorient around Laurdan fluorescent occurs (Figs. 10 B and 11 B). At high temperatures, low moiety during its excited-state lifetime. the area excitation and emission GP values characterize of In apolar solvents, such as hexane or cyclohexane, the liquid-crystalline-like phase, where dipolar relaxation oc- Laurdan shows very blue emission and excitation spectra curs. At intermediate temperature, intermediate excitation and a very low quantum yield (Parasassi et al., 1991). In and emission GP values are observed, and this area could be these apolar solvents, Laurdan fluorescence lifetime is of and characterized by the coexistence of gel liquid-crystalline the order of 100 ps (Parasassi et al., 1993). Hydrogen the domains. By increasing cholesterol concentration, plots bonding solvents cause a red shift of emission and excita- and of the individual samples are shifted to higher excitation tion spectra. Dipolar reorientation of solvent during Laur- emission GP values.

"It B "irz rz izi % IN "It rz. qz. -% r's 'N 1.1, IN Cli

I v I

~~~ ~ ~ ~ ~ ~ ~ ~ o__z

FIGURE 10 Three-dimensional surface obtained as reported in Fig. 9 from Laurdan excitation and emission GP values and from the temperature FIGURE 11 Three-dimensional surface obtained as reported in Fig. 9 of their measurement in vesicles composed of the equimolar DLPC-DPPC from Laurdan excitation and emission GP values and from the temperature mixture with cholesterol concentrations of (A): 0, 3, 5, 10, 15, 20, 25, 30, of their measurement in vesicles composed of DLPC with cholesterol con- 35, 40, 45, and 50 mol %; (B): 25, 30, 35, 40, 45, 50, 55, and 60 mol %. centrations of (A): 0, 3, 5, 10, 15, 20, 25, 30, 35, 40, 45, and 50 mol %; (B): The excitation GP values were measured using excitation at 370 nm and the 25, 30, 35, 40, 45, 50, 55, and 60 mol %. The excitation GP values were emission GP values using emission at 490 nm. Intensities of the blanks were measured using excitation at 370 nm and the emission GP values using subtracted. In both figures the individual three-dimensional curves of emission at 490 nm. In both figures the individual three-dimensional curves samples at each cholesterol concentration are superimposed. of samples at each cholesterol concentration are superimposed. Parasassi et al. Cholesterol Suppresses Phospholipid Domains 129 dan excited-state lifetime can cause an additional time- 1 to 30 mol %, and to the equimolar DLPC-DPPC mixture dependent red shift of the emission (Parasassi et al., 1990). with cholesterol concentrations from 3 to 60 mol %, the ex- If the rate of relaxation is very small, such as in the phos- citation spectrum slightly shifts toward the blue, indicating pholipid gel phase, this additional emission spectral shift is a less polar environment (Fig. 1, A and C) and in agreement small (Parasassi et al., 1993). In the phospholipid liquid- with the reported effect of cholesterol of reducing the hy- crystalline phase, relaxation processes can occur at a much dration ofbilayers (Levine and Wilkins, 1971). The emission faster rate, and we can observe a very large emission red spectrum shows a blue shift, also indicating a less polar en- shift (Parasassi et al., 1993). vironment. After the addition of cholesterol, both Laurdan One important consideration relates to the nature of the ground and excited states appear less polar. Besides the small molecule that reorients during the exited-state lifetime. We blue shift of the excitation spectrum in the pure phospholipid have demonstrated that the rate and extent of the relaxation gel phase, i.e., in the DMPC samples at low temperature, the process are independent on the nature of the phospholipid peculiar behavior of the excitation red band, centered at polar residue and are also independent on the pH value of the about 390 nm, must be described. In apolar solvents, the blue medium (Parasassi et al., 1991). This observation rules out shift of the excitation spectrum is also accompanied by the the motion of phospholipid polar heads as the molecular en- decrease of this red band. By adding cholesterol to the gel tity responsible for reorientation. Measurements of the po- phase, this red band also slightly decreases in its relative larization value across the emission band also rules out the intensity, in agreement with a decrease ofpolarity of Laurdan possibility that the Laurdan molecule itself rotates to better environment. Variations of the ratio between the two exci- orient with respect to dipoles or charges of the membrane tation bands are well represented by the emission GP value. polar surface (Parasassi et al., 1990). To rule out a further The decreased polarity of Laurdan environment after cho- possibility that residues of Laurdan molecules rotate and lesterol addition is clearly displayed by the decrease of the reorient during the probe excited state, we performed emission GP value at low temperatures (Figs. 2 B and 3 B). measurements on Laurdan analogues. Our results (Parasassi In this case, the small decrease in the intensity of the exci- and Gratton, unpublished observations) excluded the hypoth- tation band at 410 nm is not indicative of a more liquid- esis that this intermolecular relaxation was the source of the crystalline-like environment of Laurdan fluorescent moiety, relaxation process. Instead, the possibility exists that a few but only of a reduced stabilization of the ground state Lay water molecules present in the hydrophilic-hydrophobic in- population which is typical ofLaurdan molecules surrounded terface of the bilayer, close to the naphthalene moiety of by dipoles with a favorable orientation. The same blue shift Laurdan, can reorient during Laurdan excited-state lifetime. ofboth Laurdan excitation and emission spectra are observed The fluorescent moiety of Laurdan is located at the level of in the equimolar DLPC-DPPC mixture (Fig. 1 C). In this the phospholipid glycerol backbone. The dynamic properties mixture, the emission GP value slightly increases instead of of these water molecules in the membrane are reduced with decreasing as in the DMPC sample (Fig. 3 D). In the case respect to the bulk water molecules of the solvent. The ori- of the mixture of the two phospholipids, the decrease in po- entational rate must be on the order of nanoseconds in the larity induced by cholesterol is masked by the dipolar re- phospholipid liquid-crystalline phase. We have proposed laxation, even in small amounts, due to the presence ofDLPC that a few water molecules present in the bilayer close to with a Tm of -0C. Laurdan naphthalene moiety are the molecular entities that produce the dipolar relaxation process responsible for the red emission spectral shift displayed by Laurdan (Parasassi et al., Laurdan fluorescence spectra in the 1991). In the gel phase this process is strongly reduced either liquid-crystalline phase and in the presence of because of the absence of water or because the phospholipid cholesterol a in the tight packing prevents rapid relaxation, nanosecond Cholesterol addition to DLPC vesicles, to the liquid- of scale, water molecules. crystalline phase of DMPC vesicles, and to the equimolar A peculiar feature of Laurdan excitation spectrum is re- DLPC-DPPC mixture has a larger effect than on the gel lated to the red excitation band, the intensity of which in- phase. In this phase state the emission spectrum shows a blue creases in solvents and in polar hydrogen bonding solvents. shift. Both the decreased polarity at the level of the glycerol This effect has been attributed to the stabilization of the La backbone of phospholipids in the bilayer and the ordering conformation In transitions. phospholipids, this red excita- effect of cholesterol (Vist and Davis, 1990; et tion band Ipsen al., corresponds also to a blue-emitting population, 1987), that renders the acyl chains conformation in the bi- intense in especially gel phase phospholipids (Parasassi et al., layers similar to that of the gel phase, reduce the amount of 1990), where very little relaxation occurs. dipolar relaxation and cause a blue shift of the emission spec- trum, similar to that observed in gel phase phospholipids. On Laurdan fluorescence spectra in gel phase and in the contrary, the excitation spectrum shows a red shift, with the presence of cholesterol a relevant increase of the red excitation band (Fig. 1, B and D). After addition of cholesterol, the prevailing effect on the At low temperature, when cholesterol is added to DMPC in excitation spectrum is due to the decreased amount of re- the gel phase, at all the concentrations used in this work, from laxation rather than to decreased polarity that should cause 130 Biophysical Journal Volume 66 January 1994 a blue shift. Laurdan excitation spectrum seems quite sen- 340 nm at all emission wavelengths (Eq. 2). Excitation at 340 sitive to the ordering effect of cholesterol on the acyl chain nm slightly selects Laurdan molecules with energetically un- conformation. Modifications of Laurdan emission due to favorable orientation of the surrounding dipoles (Parasassi cholesterol are also quite relevant, both the reduced polarity et al., 1990), only a part of which will relax during the probe and the increased conformational order contributing to the excited state. Excitation at 410 nm preferentially selects blue shift of the emission. Laurdan molecules mostly surrounded by favorably oriented At low temperature and when cholesterol is present, the dipoles, which will show a relaxed emission. Consequently, variations of Laurdan emission and excitation spectra in by observing the red emission at 490 nm, higher intensity will phospholipids are largely due to the decrease of polarity, originate from the 410 nm band of excitation. By moving the probably because of the decreased water concentration at the emission toward the blue, the intensity arising from 410 nm level of the fluorescent moiety of Laurdan molecules in the excitation will decrease, and the GP value will also decrease bilayer (Levine and Wilkins, 1971), with the consequence of (Parasassi et al., 1993). Again, this trend can only be ob- a further decrease of the dipolar relaxation and of a decreased served in a homogeneous liquid-crystalline phase in the pres- stabilization of the L. conformations. In DLPC at all tem- ence of the dipolar relaxation process (Figs. 4 and 5, spectra peratures, in DMPC, and in the equimolar DLPC-DPPC mix- at 650C). ture at high temperatures, where dipolar relaxation prevails, The coexistence of gel and liquid-crystalline phases, in the major consequence of cholesterol addition is the reduc- the absence of cholesterol, gives rise to an opposite wave- tion of the molecular motion of water molecules rather than length dependence of the excitation and emission GP spec- decreasing the water concentration itself, thus rendering tra, as previously discussed (Parasassi et al., 1993). Laurdan Laurdan spectra closer to those observed in gel phase phos- molecules, in a domain of phospholipids in gel phase, pref- pholipids where very little relaxation ofwater dipoles occurs. erentially absorb in the 390 nm region and emit in the 440 nm region. Laurdan molecules in a domain of liquid- Coexisting gel and liquid-crystalline phospholipid crystalline phase preferentially absorb in the 360 nm region domains and emit in the 490 nm region. In the presence of both co- existing phases (sample of equimolar mixture of DLPC and In vesicles composed of the equimolar DLPC-DPPC mixture DPPC, at temperatures between 10 and 300C) the excitation and at temperatures between 10 and 30°C, Laurdan spectra, GP spectrum should then decrease toward shorter wave- which are intermediate between those characteristic of the lengths, while the emission GP spectrum should increase. gel and of the liquid-crystalline phase, are observed. In this Therefore, phase domains coexistence has an opposite trend temperature range, the above phospholipid mixture is known with respect to that observed only in the presence of the to display coexistence of gel and liquid-crystalline domains with the excitation GP spectrum (Shimshick and McConnel, 1973; Parasassi et al., 1984). The dipolar relaxation, show- effect of increasing cholesterol concentration in this sample ing increasing values with the increase of the excitation is similar to that observed in pure liquid-crystalline samples. wavelength, and with the emission GP spectrum showing The Laurdan emission spectrum is blue-shifted and the ex- decreasing values with the increase of the emission wave- citation spectrum shows the increase in intensity of the red length (Parasassi et al., 1990, 1991, 1993) (Fig. 6). Finally, band. in the homogeneous gel phase, no appreciable wavelength dependence of excitation and emission GP values is ob- served (Figs. 4 and 5, spectra at 10C). Laurdan excitation and emission GP in the The behavior ofthe GP spectra as a function ofwavelength presence of cholesterol in the presence of coexisting domains is modified by the To explain the wavelength dependence of the excitation and presence of cholesterol, and the type and amount of modi- emission GP value, we first analyzed the expected behavior fication depends on cholesterol concentration. In the of a homogeneous liquid-crystalline phase in the absence of equimolar DLPC-DPPC mixture, at low cholesterol concen- cholesterol. It is characteristic of the dipolar relaxation pro- trations up to about 10 mol %, the wavelength dependence cess that by exciting in the blue part of the spectrum we of excitation and emission GP is similar to that observed photoselect those molecules with an energetically unfavor- without cholesterol. Differences are observed only for the able ground state, i.e., those molecules surrounded by ran- temperature at which the wavelength dependence of exci- domly oriented dipoles. As a consequence, by moving the tation and emission GP typical of the coexistence of both excitation toward the blue, more blue emitting molecules are phases can be observed. In correspondence to the shift of the excited. The excitation GP value, which depends on the dif- phase transition temperature reported in Figs. 2 and 3, emis- ference between the emission intensities at 440 and 490 nm sion and excitation GP spectra with the characteristic wave- (Eq. 3), should increase toward shorter excitation wave- length dependence observed when phase domains coexist are lengths. Of course, this effect is only visible in a homoge- observed at temperatures progressively higher with the in- neous liquid-crystalline phase where dipolar relaxation oc- crease of cholesterol concentration (Fig. 6). At cholesterol curs (Figs. 4 and 5, spectra at 65°C). concentrations higher than 10 mol %, the above described The emission GP spectra are obtained by measuring the behavior of GP values is no longer observed. From low to difference between the excitation intensities at 410 nm and high temperatures both excitation and emission GP spectra Parasassi et al. Cholesterol Suppresses Phospholipid Domains 131 pass from a wavelength dependence typical of the pure gel substitution of water molecules with the hydroxyl residue of to that typical of the pure liquid-crystalline phase. No in- cholesterol. Instead, the increase in the phospholipid acyl termediate behavior indicative of gel and liquid-crystalline chain conformational order or a tight packing of the bilayer domain coexistence was observed (Fig. 6). close to that of the gel phase must be invoked. Since the excitation spectrum shows opposite response to the polarity or to the order, while the emission spectrum is shifted to the The absence of rapid mixing same direction by modification of these two quantities, Our experiments failed to detect separate spectroscopic sig- the emission GP values do not vary of the same amount as nals in the regions in which liquid phase or gel phase im- the excitation GP. Indeed, by plotting the excitation GP val- miscibility has been postulated. Separated and resolvable ues versus the emission GP values obtained at each tem- Laurdan spectroscopic properties indicative of the coexist- perature, the curves do not overlap. Using the temperature as ence ofdomains with different polarity or different molecular the third axis, a surface is obtained where different areas can order could not be seen. As we noted before, the same failure be characterized (Figs. 9-11). High values of both excitation occurred for NMR and ESR experiments (Vist and Davis, and emission GP at low temperature characterize an area of 1990). Since the time scale of the fluorescent phenomenon gel-like phase. Low excitation and emission GP values at is on the order of 10-' s, if an averaging of emission prop- high temperature characterize an area where a large amount erties should be invoked to explain the absence of distinct of dipolar relaxation occurs, such as in the liquid-crystalline signals, this argument set a limit on the size and lifetime of phase. The intermediate areas characterize the possible co- domains. The lifetime of the domain should be shorter that existence of separate phases. Noticeably, the sharp tempera- 5 ns and the size should be no larger than 50 lipid molecules, ture range of DMPC phase transition is well enclosed in the based on an estimate of the diffusion rate of small molecules intermediate area of phase coexistence, while at 30 mol % in the membrane plane (Parasassi and Gratton, 1992). Similar cholesterol in DMPC the liquid-crystalline valley is never estimates on the dimension of domains in vesicles composed reached, even at 65°C (Fig. 9). Similarly, plots of DLPC of equimolar mixtures of gel and liquid-crystalline phases, vesicles with cholesterol concentrations 230 mol % are not without cholesterol, were obtained on the basis of time- contained in the liquid-crystalline area. resolved measurements of Laurdan emission (Parasassi et al., A further intermediate area can be observed in the sur- 1993). From the diffusion coefficient in the two phases and face obtained using the equimolar DLPC-DPPC mixture, from the rate of interconversion of Laurdan between the where a peak is present, at intermediate temperature, from phases, an average area of 20-50 A2 has been calculated for about 10°C to about 30°C (Fig. 10 A). Nevertheless, when the domains (Parasassi et al., 1993). It is quite puzzling that the surface is obtained with the data from the same phos- at least three spectroscopic techniques with quite different pholipid mixture but in the presence of cholesterol concen- characteristic time scales (NMR, ESR, and fluorescence) trations from 25 to 60 mol %, the intermediate peak disap- have failed to detect distinct environments in a region of pears (Fig. 10 B). phase diagram where separated domains are expected to ex- From Figs. 10 B and 11 B we can observe that at cho- ist. The possibility remains that these theoretical domains are lesterol concentrations 230 mol % a plateau is reached very small, composed of a few phospholipid molecules, and that the surface is flat. The concentration of choles- and/or fluctuating more rapidly than previously anticipated. terol in most biological membranes is 230 mol %. Both Moreover, the different Laurdan spectroscopic properties, from our results and from the reported phase diagrams specific for each phase domain (Parasassi et al., 1991 and (Vist and Davis, 1990; Sankaram and Thompson, 1990b; 1993), are progressively closer by increasing cholesterol Tampe et al., 1991), this is the region of homogeneous concentration. In the case of pure phospholipid separate do- liquid-ordered phase. Different concentrations of choles- mains, the results showed that each phase is profoundly terol affect differently Laurdan dipolar relaxation and its modified by the presence of the other (Parasassi et al., 1993). response to polarity, so that instead of overlapping curves By adding cholesterol to phospholipid mixtures showing do- a surface can be obtained. main segregation, as in the present work, the modification of At low cholesterol concentration, the intermediate area each phase is even more significant, so that separate spec- between the gel and the liquid-crystalline regions can be troscopic Laurdan parameters cannot be distinguished. representative both of coexisting domains of liquid-ordered and liquid-disordered phases, i.e., liquid-phase immiscibil- Analysis of the GP excitation-emission surfaces ity, and of a homogeneous environment ofintermediate prop- erties. Our analysis based on the wavelength dependence of In general, from our results we can conclude that Laurdan excitation and emission GP values rules out the possibility sensitivity to the polarity and to the kinetics of dipolar re- of coexisting domains, as reported above. Instead, evidence laxation of its environment detects a decreased polarity and of solid-phase immiscibility can be found in already reported a lower amount of relaxation in phospholipids after addition oxygen quenching experiments (Parasassi and Gratton, of cholesterol. The decrease of the amount of dipolar relax- 1992) where 10 mol % cholesterol was added to gel phase ation produced by cholesterol in the liquid-crystalline phase phospholipids. In this case a biphasic behavior of the Stem- is too large to be attributed to a decreased polarity, i.e., to the Volmer plot was observed, indicative of different Laurdan 132 Biophysical Journal Volume 66 January 1994 environments. However, when experiments are performed in Parasassi, T., G. De Stasio, A. d'Ubaldo, and E. Gratton. 1990. Phase fluc- tuation in phospholipid membranes revealed by Laurdan fluorescence. the liquid-crystalline phase with various cholesterol concen- Biophys. J. 57:1179-1186 trations linear Stern-Volmer plots were always obtained, Parasassi, T., G. De Stasio, G. Ravagnan, R. M. Rusch, and E. Gratton. 1991. with no indications of liquid-phase immiscibility (Fig. 8). Quantitation of lipid phases in phospholipid vesicles by the generalized polarization of Laurdan fluorescence. Biophys. J. 60:179-180. Parasassi, T., M. Di Stefano, G. Ravagnan, 0. Sapora, and E. Gratton. 1992. This work was supported by Consiglio Nazionale Ricerche (TP, MDS, ML, Membrane aging during cells growth ascertained by Laurdan generalized GR) and by the National Institutes of Health (RR03155) (EG). polarization. Exp. Cell Res. 202:432-439. Parasassi, T., and E. Gratton, 1992. Packing of phospholipid vesicles studied by oxygen quenching of Laurdan fluorescence. J. Fluorescence. 2: REFERENCES 167-174. Parasassi, T., G. Ravagnan, R. M. Rusch, and E. Gratton. 1993. Modulation Ben-Yasar, V., and Y. Barenholz. 1989. The interaction of cholesterol and and dynamics of phase properties in phospholipid mixtures detected by cholest-4-en-3-one with dipalmitoylphosphatidyl-choline. Comparison Laurdan fluorescence. Photochem. Photobiol. 57:403-410. based on the use of three fluorophores. Biochim. Biophys. Acta. 985: 271-278. Rubenstein, J. L., B. A. Smith, and H. D. McConnell. 1979. Lateral diffusion Ipsen, J. H., G. Karlstrom, 0. G. Mouritsen, H. Wennerstrom, and M. H. in binary mixtures of cholesterol and phosphatidylcholines. Proc. Natl. Zuckermann. 1987. Phase equilibria in the phosphatidylcholine- Acad. Sci. USA. 76:15-18. cholesterol system. Biochim. Biophys. Acta. 905:162-172. Sankaram, M. B., and T. E. Thompson. 1990a. Modulation of phospholipid Keough, K. M, B. Giffin, and P. L. Matthews. 1989. Phosphatidylcholine- acyl chain order by cholesterol. A solid-state 2H nuclear magnetic reso- cholesterol interactions: bilayers of heteroacid lipids containing linoleate nance study. Biochemistry. 29:10676-10684. lose calorimetric transitions at low cholesterol concentration. Biochim. Sankaram, M. B., and T. E. Thompson. 1990b. Interaction of cholesterol Biophys. Acta. 983:51-55. with various glycerophospholipids and sphingomyelin. Biochemistry. 29: Lakowicz, J. R., and G. Weber. 1973. Quenching offluorescence by oxygen. 10670-10675. A probe for structural fluctuations in macromolecules. Biochemistry. 12: Schroeder, F., J. R. Jefferson, A. B. Kier, J. Knittel, T. J. Scallen, W. Gibson 4161-4170. Wood, and I. Hapala. 1991. Membrane cholesterol dynamics: cholesterol Levine, Y. K., and M. H. F. Wilkins. 1971. Structure of oriented lipid domains and kinetic pools. Proc. Soc. Exp. Biol. Med. 196:235-252. bilayers. Nature New Bio. 230:69-72. Shimshick, E. J., and H. M. McConnel. 1973. Lateral phase separation in Melchior, D. L., F. J. Scavitto, and J. M. Steim. 1980. Dilatometry of phospholipid membranes. Biochemistry. 12:2351-2360. dipalmitoyl-lecithin-cholesterol bilayers. Biochemistry. 19:4828-4834. Subczynski, W. K., W. E. Antholine, J. S. Hyde, and A. Kusumi. 1990. Mortensen, N. K., W. Pfeiffer, E. Sackmann, and W. Knoll. 1988. Structural Microimmiscibility and three-dimensional dynamic structure of properties of a phosphatidylcholine-cholesterol system as studied by phosphatidylcholine-cholesterol membranes: translational diffusion of a small-angle neutron scattering: ripple structure and phase diagram. copper complex in the membrane. Biochemistry. 29:7936-7945. Biochim. Biophys. Acta. 945:221-245. Tampe, R., A. von Lukas, and H. J. Galla. 1991. Glycophorin-induced Mouritsen, 0. G. 1991. Theoretical models of phospholipid phase transi- cholesterol-phospholipid domains in dimyristoylphosphatidylcholine bi- tions. Chem. Phys. Lipids. 57:179-194. layer vesicles. Biochemistry. 30:4909-4916. Nemecz, G., and F. Shroeder. 1988. Time-resolved fluorescence investi- van Langen, H., G. van Ginkel, D. Shaw, and Y. K. Levine. 1989. The gation on membrane cholesterol heterogeneity and exchange. Biochem- fidelity response by 1-[4-(trimethylammonio)phenyl]-6-phenyl-1,3,5- istry. 27:7740-7749. hexatriene in time-resolved fluorescence anisotropy measurements on Parasassi, T., F. Conti, M. Glaser, and E. Gratton. 1984. Detection of lipid vesicles. Eur. Biophys. J. 17:37-48. phospholipid phase separation. A multifrequency phase fluorimetry Vist, M. R., and J. D. Davis. 1990. Phase equilibria of cholesterol/ study of 1:6-diphenyl-1,3,5-hexatriene fluorescence. J. Bio. Chem. dipalmitoyl-phosphatidylcholine mixtures; 2H nuclear magnetic reso- 259:14011-14017. nance and differential scanning calorimetry. Biochemistry. 29:451-464.